rdb.c 32.8 KB
Newer Older
1 2 3 4
#include "redis.h"
#include "lzf.h"    /* LZF compression library */

#include <math.h>
5 6 7 8 9
#include <sys/types.h>
#include <sys/time.h>
#include <sys/resource.h>
#include <sys/wait.h>
#include <arpa/inet.h>
10
#include <sys/stat.h>
11

12 13 14 15 16 17 18 19 20
/* Convenience wrapper around fwrite, that returns the number of bytes written
 * to the file instead of the number of objects (see fwrite(3)) and -1 in the
 * case of an error. It also supports a NULL *fp to skip writing altogether
 * instead of writing to /dev/null. */
static int rdbWriteRaw(FILE *fp, void *p, size_t len) {
    if (fp != NULL && fwrite(p,len,1,fp) == 0) return -1;
    return len;
}

21
int rdbSaveType(FILE *fp, unsigned char type) {
22
    return rdbWriteRaw(fp,&type,1);
23 24 25 26
}

int rdbSaveTime(FILE *fp, time_t t) {
    int32_t t32 = (int32_t) t;
27
    return rdbWriteRaw(fp,&t32,4);
28 29 30 31 32
}

/* check rdbLoadLen() comments for more info */
int rdbSaveLen(FILE *fp, uint32_t len) {
    unsigned char buf[2];
33
    int nwritten;
34 35 36 37

    if (len < (1<<6)) {
        /* Save a 6 bit len */
        buf[0] = (len&0xFF)|(REDIS_RDB_6BITLEN<<6);
38
        if (rdbWriteRaw(fp,buf,1) == -1) return -1;
39
        nwritten = 1;
40 41 42 43
    } else if (len < (1<<14)) {
        /* Save a 14 bit len */
        buf[0] = ((len>>8)&0xFF)|(REDIS_RDB_14BITLEN<<6);
        buf[1] = len&0xFF;
44
        if (rdbWriteRaw(fp,buf,2) == -1) return -1;
45
        nwritten = 2;
46 47 48
    } else {
        /* Save a 32 bit len */
        buf[0] = (REDIS_RDB_32BITLEN<<6);
49
        if (rdbWriteRaw(fp,buf,1) == -1) return -1;
50
        len = htonl(len);
51
        if (rdbWriteRaw(fp,&len,4) == -1) return -1;
52
        nwritten = 1+4;
53
    }
54
    return nwritten;
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
}

/* Encode 'value' as an integer if possible (if integer will fit the
 * supported range). If the function sucessful encoded the integer
 * then the (up to 5 bytes) encoded representation is written in the
 * string pointed by 'enc' and the length is returned. Otherwise
 * 0 is returned. */
int rdbEncodeInteger(long long value, unsigned char *enc) {
    /* Finally check if it fits in our ranges */
    if (value >= -(1<<7) && value <= (1<<7)-1) {
        enc[0] = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_INT8;
        enc[1] = value&0xFF;
        return 2;
    } else if (value >= -(1<<15) && value <= (1<<15)-1) {
        enc[0] = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_INT16;
        enc[1] = value&0xFF;
        enc[2] = (value>>8)&0xFF;
        return 3;
    } else if (value >= -((long long)1<<31) && value <= ((long long)1<<31)-1) {
        enc[0] = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_INT32;
        enc[1] = value&0xFF;
        enc[2] = (value>>8)&0xFF;
        enc[3] = (value>>16)&0xFF;
        enc[4] = (value>>24)&0xFF;
        return 5;
    } else {
        return 0;
    }
}

/* String objects in the form "2391" "-100" without any space and with a
 * range of values that can fit in an 8, 16 or 32 bit signed value can be
 * encoded as integers to save space */
int rdbTryIntegerEncoding(char *s, size_t len, unsigned char *enc) {
    long long value;
    char *endptr, buf[32];

    /* Check if it's possible to encode this value as a number */
    value = strtoll(s, &endptr, 10);
    if (endptr[0] != '\0') return 0;
    ll2string(buf,32,value);

    /* If the number converted back into a string is not identical
     * then it's not possible to encode the string as integer */
    if (strlen(buf) != len || memcmp(buf,s,len)) return 0;

    return rdbEncodeInteger(value,enc);
}

int rdbSaveLzfStringObject(FILE *fp, unsigned char *s, size_t len) {
    size_t comprlen, outlen;
    unsigned char byte;
107
    int n, nwritten = 0;
108 109 110 111 112 113 114 115 116 117 118 119 120
    void *out;

    /* We require at least four bytes compression for this to be worth it */
    if (len <= 4) return 0;
    outlen = len-4;
    if ((out = zmalloc(outlen+1)) == NULL) return 0;
    comprlen = lzf_compress(s, len, out, outlen);
    if (comprlen == 0) {
        zfree(out);
        return 0;
    }
    /* Data compressed! Let's save it on disk */
    byte = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_LZF;
121 122
    if ((n = rdbWriteRaw(fp,&byte,1)) == -1) goto writeerr;
    nwritten += n;
123 124 125 126 127 128 129

    if ((n = rdbSaveLen(fp,comprlen)) == -1) goto writeerr;
    nwritten += n;

    if ((n = rdbSaveLen(fp,len)) == -1) goto writeerr;
    nwritten += n;

130 131
    if ((n = rdbWriteRaw(fp,out,comprlen)) == -1) goto writeerr;
    nwritten += n;
132

133
    zfree(out);
134
    return nwritten;
135 136 137 138 139 140 141

writeerr:
    zfree(out);
    return -1;
}

/* Save a string objet as [len][data] on disk. If the object is a string
142
 * representation of an integer value we try to save it in a special form */
143 144
int rdbSaveRawString(FILE *fp, unsigned char *s, size_t len) {
    int enclen;
145
    int n, nwritten = 0;
146 147 148 149 150

    /* Try integer encoding */
    if (len <= 11) {
        unsigned char buf[5];
        if ((enclen = rdbTryIntegerEncoding((char*)s,len,buf)) > 0) {
151
            if (rdbWriteRaw(fp,buf,enclen) == -1) return -1;
152
            return enclen;
153 154 155 156 157 158
        }
    }

    /* Try LZF compression - under 20 bytes it's unable to compress even
     * aaaaaaaaaaaaaaaaaa so skip it */
    if (server.rdbcompression && len > 20) {
159 160 161 162
        n = rdbSaveLzfStringObject(fp,s,len);
        if (n == -1) return -1;
        if (n > 0) return n;
        /* Return value of 0 means data can't be compressed, save the old way */
163 164 165
    }

    /* Store verbatim */
166 167 168
    if ((n = rdbSaveLen(fp,len)) == -1) return -1;
    nwritten += n;
    if (len > 0) {
169
        if (rdbWriteRaw(fp,s,len) == -1) return -1;
170 171 172
        nwritten += len;
    }
    return nwritten;
173 174 175 176 177
}

/* Save a long long value as either an encoded string or a string. */
int rdbSaveLongLongAsStringObject(FILE *fp, long long value) {
    unsigned char buf[32];
178
    int n, nwritten = 0;
179 180
    int enclen = rdbEncodeInteger(value,buf);
    if (enclen > 0) {
181
        return rdbWriteRaw(fp,buf,enclen);
182 183 184 185
    } else {
        /* Encode as string */
        enclen = ll2string((char*)buf,32,value);
        redisAssert(enclen < 32);
186 187
        if ((n = rdbSaveLen(fp,enclen)) == -1) return -1;
        nwritten += n;
188 189
        if ((n = rdbWriteRaw(fp,buf,enclen)) == -1) return -1;
        nwritten += n;
190
    }
191
    return nwritten;
192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244
}

/* Like rdbSaveStringObjectRaw() but handle encoded objects */
int rdbSaveStringObject(FILE *fp, robj *obj) {
    /* Avoid to decode the object, then encode it again, if the
     * object is alrady integer encoded. */
    if (obj->encoding == REDIS_ENCODING_INT) {
        return rdbSaveLongLongAsStringObject(fp,(long)obj->ptr);
    } else {
        redisAssert(obj->encoding == REDIS_ENCODING_RAW);
        return rdbSaveRawString(fp,obj->ptr,sdslen(obj->ptr));
    }
}

/* Save a double value. Doubles are saved as strings prefixed by an unsigned
 * 8 bit integer specifing the length of the representation.
 * This 8 bit integer has special values in order to specify the following
 * conditions:
 * 253: not a number
 * 254: + inf
 * 255: - inf
 */
int rdbSaveDoubleValue(FILE *fp, double val) {
    unsigned char buf[128];
    int len;

    if (isnan(val)) {
        buf[0] = 253;
        len = 1;
    } else if (!isfinite(val)) {
        len = 1;
        buf[0] = (val < 0) ? 255 : 254;
    } else {
#if (DBL_MANT_DIG >= 52) && (LLONG_MAX == 0x7fffffffffffffffLL)
        /* Check if the float is in a safe range to be casted into a
         * long long. We are assuming that long long is 64 bit here.
         * Also we are assuming that there are no implementations around where
         * double has precision < 52 bit.
         *
         * Under this assumptions we test if a double is inside an interval
         * where casting to long long is safe. Then using two castings we
         * make sure the decimal part is zero. If all this is true we use
         * integer printing function that is much faster. */
        double min = -4503599627370495; /* (2^52)-1 */
        double max = 4503599627370496; /* -(2^52) */
        if (val > min && val < max && val == ((double)((long long)val)))
            ll2string((char*)buf+1,sizeof(buf),(long long)val);
        else
#endif
            snprintf((char*)buf+1,sizeof(buf)-1,"%.17g",val);
        buf[0] = strlen((char*)buf+1);
        len = buf[0]+1;
    }
245
    return rdbWriteRaw(fp,buf,len);
246 247 248 249
}

/* Save a Redis object. */
int rdbSaveObject(FILE *fp, robj *o) {
250 251
    int n, nwritten = 0;

252 253
    if (o->type == REDIS_STRING) {
        /* Save a string value */
254 255
        if ((n = rdbSaveStringObject(fp,o)) == -1) return -1;
        nwritten += n;
256 257 258 259 260 261 262 263
    } else if (o->type == REDIS_LIST) {
        /* Save a list value */
        if (o->encoding == REDIS_ENCODING_ZIPLIST) {
            unsigned char *p;
            unsigned char *vstr;
            unsigned int vlen;
            long long vlong;

264 265 266
            if ((n = rdbSaveLen(fp,ziplistLen(o->ptr))) == -1) return -1;
            nwritten += n;

267 268 269
            p = ziplistIndex(o->ptr,0);
            while(ziplistGet(p,&vstr,&vlen,&vlong)) {
                if (vstr) {
270
                    if ((n = rdbSaveRawString(fp,vstr,vlen)) == -1)
271
                        return -1;
272
                    nwritten += n;
273
                } else {
274
                    if ((n = rdbSaveLongLongAsStringObject(fp,vlong)) == -1)
275
                        return -1;
276
                    nwritten += n;
277 278 279 280 281 282 283 284
                }
                p = ziplistNext(o->ptr,p);
            }
        } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
            list *list = o->ptr;
            listIter li;
            listNode *ln;

285 286 287
            if ((n = rdbSaveLen(fp,listLength(list))) == -1) return -1;
            nwritten += n;

288 289 290
            listRewind(list,&li);
            while((ln = listNext(&li))) {
                robj *eleobj = listNodeValue(ln);
291 292
                if ((n = rdbSaveStringObject(fp,eleobj)) == -1) return -1;
                nwritten += n;
293 294 295 296 297 298
            }
        } else {
            redisPanic("Unknown list encoding");
        }
    } else if (o->type == REDIS_SET) {
        /* Save a set value */
299 300 301 302
        if (o->encoding == REDIS_ENCODING_HT) {
            dict *set = o->ptr;
            dictIterator *di = dictGetIterator(set);
            dictEntry *de;
303

304 305 306
            if ((n = rdbSaveLen(fp,dictSize(set))) == -1) return -1;
            nwritten += n;

307 308
            while((de = dictNext(di)) != NULL) {
                robj *eleobj = dictGetEntryKey(de);
309 310
                if ((n = rdbSaveStringObject(fp,eleobj)) == -1) return -1;
                nwritten += n;
311 312 313 314
            }
            dictReleaseIterator(di);
        } else if (o->encoding == REDIS_ENCODING_INTSET) {
            intset *is = o->ptr;
315
            int64_t llval;
316 317
            int i = 0;

318 319 320
            if ((n = rdbSaveLen(fp,intsetLen(is))) == -1) return -1;
            nwritten += n;

321
            while(intsetGet(is,i++,&llval)) {
322 323
                if ((n = rdbSaveLongLongAsStringObject(fp,llval)) == -1) return -1;
                nwritten += n;
324 325 326
            }
        } else {
            redisPanic("Unknown set encoding");
327 328 329 330 331 332 333
        }
    } else if (o->type == REDIS_ZSET) {
        /* Save a set value */
        zset *zs = o->ptr;
        dictIterator *di = dictGetIterator(zs->dict);
        dictEntry *de;

334 335 336
        if ((n = rdbSaveLen(fp,dictSize(zs->dict))) == -1) return -1;
        nwritten += n;

337 338 339 340
        while((de = dictNext(di)) != NULL) {
            robj *eleobj = dictGetEntryKey(de);
            double *score = dictGetEntryVal(de);

341 342 343 344
            if ((n = rdbSaveStringObject(fp,eleobj)) == -1) return -1;
            nwritten += n;
            if ((n = rdbSaveDoubleValue(fp,*score)) == -1) return -1;
            nwritten += n;
345 346 347 348 349
        }
        dictReleaseIterator(di);
    } else if (o->type == REDIS_HASH) {
        /* Save a hash value */
        if (o->encoding == REDIS_ENCODING_ZIPMAP) {
350
            size_t l = zipmapBlobLen((unsigned char*)o->ptr);
351

352
            if ((n = rdbSaveRawString(fp,o->ptr,l)) == -1) return -1;
353
            nwritten += n;
354 355 356 357
        } else {
            dictIterator *di = dictGetIterator(o->ptr);
            dictEntry *de;

358 359 360
            if ((n = rdbSaveLen(fp,dictSize((dict*)o->ptr))) == -1) return -1;
            nwritten += n;

361 362 363 364
            while((de = dictNext(di)) != NULL) {
                robj *key = dictGetEntryKey(de);
                robj *val = dictGetEntryVal(de);

365 366 367 368
                if ((n = rdbSaveStringObject(fp,key)) == -1) return -1;
                nwritten += n;
                if ((n = rdbSaveStringObject(fp,val)) == -1) return -1;
                nwritten += n;
369 370 371 372 373 374
            }
            dictReleaseIterator(di);
        }
    } else {
        redisPanic("Unknown object type");
    }
375
    return nwritten;
376 377 378 379 380 381
}

/* Return the length the object will have on disk if saved with
 * the rdbSaveObject() function. Currently we use a trick to get
 * this length with very little changes to the code. In the future
 * we could switch to a faster solution. */
382 383 384 385
off_t rdbSavedObjectLen(robj *o) {
    int len = rdbSaveObject(NULL,o);
    redisAssert(len != -1);
    return len;
386 387
}

388 389 390 391
/* Save a key-value pair, with expire time, type, key, value.
 * On error -1 is returned.
 * On success if the key was actaully saved 1 is returned, otherwise 0
 * is returned (the key was already expired). */
392 393
int rdbSaveKeyValuePair(FILE *fp, robj *key, robj *val,
                        time_t expiretime, time_t now)
394
{
395 396
    int vtype;

397 398 399 400 401 402 403
    /* Save the expire time */
    if (expiretime != -1) {
        /* If this key is already expired skip it */
        if (expiretime < now) return 0;
        if (rdbSaveType(fp,REDIS_EXPIRETIME) == -1) return -1;
        if (rdbSaveTime(fp,expiretime) == -1) return -1;
    }
404 405 406 407 408
    /* Fix the object type if needed, to support saving zipmaps, ziplists,
     * and intsets, directly as blobs of bytes: they are already serialized. */
    vtype = val->type;
    if (vtype == REDIS_HASH && val->encoding == REDIS_ENCODING_ZIPMAP)
        vtype = REDIS_HASH_ZIPMAP;
409 410
    /* Save type, key, value */
    if (rdbSaveType(fp,val->type) == -1) return -1;
411
    if (rdbSaveStringObject(fp,key) == -1) return -1;
412 413 414 415
    if (rdbSaveObject(fp,val) == -1) return -1;
    return 1;
}

416 417 418 419 420 421 422 423 424
/* Save the DB on disk. Return REDIS_ERR on error, REDIS_OK on success */
int rdbSave(char *filename) {
    dictIterator *di = NULL;
    dictEntry *de;
    FILE *fp;
    char tmpfile[256];
    int j;
    time_t now = time(NULL);

A
antirez 已提交
425 426 427 428
    if (server.ds_enabled) {
        cacheForcePointInTime();
        return dsRdbSave(filename);
    }
429 430 431 432

    snprintf(tmpfile,256,"temp-%d.rdb", (int) getpid());
    fp = fopen(tmpfile,"w");
    if (!fp) {
A
antirez 已提交
433 434
        redisLog(REDIS_WARNING, "Failed opening .rdb for saving: %s",
            strerror(errno));
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455
        return REDIS_ERR;
    }
    if (fwrite("REDIS0001",9,1,fp) == 0) goto werr;
    for (j = 0; j < server.dbnum; j++) {
        redisDb *db = server.db+j;
        dict *d = db->dict;
        if (dictSize(d) == 0) continue;
        di = dictGetIterator(d);
        if (!di) {
            fclose(fp);
            return REDIS_ERR;
        }

        /* Write the SELECT DB opcode */
        if (rdbSaveType(fp,REDIS_SELECTDB) == -1) goto werr;
        if (rdbSaveLen(fp,j) == -1) goto werr;

        /* Iterate this DB writing every entry */
        while((de = dictNext(di)) != NULL) {
            sds keystr = dictGetEntryKey(de);
            robj key, *o = dictGetEntryVal(de);
456
            time_t expire;
457 458
            
            initStaticStringObject(key,keystr);
459 460
            expire = getExpire(db,&key);
            if (rdbSaveKeyValuePair(fp,&key,o,expire,now) == -1) goto werr;
461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494
        }
        dictReleaseIterator(di);
    }
    /* EOF opcode */
    if (rdbSaveType(fp,REDIS_EOF) == -1) goto werr;

    /* Make sure data will not remain on the OS's output buffers */
    fflush(fp);
    fsync(fileno(fp));
    fclose(fp);

    /* Use RENAME to make sure the DB file is changed atomically only
     * if the generate DB file is ok. */
    if (rename(tmpfile,filename) == -1) {
        redisLog(REDIS_WARNING,"Error moving temp DB file on the final destination: %s", strerror(errno));
        unlink(tmpfile);
        return REDIS_ERR;
    }
    redisLog(REDIS_NOTICE,"DB saved on disk");
    server.dirty = 0;
    server.lastsave = time(NULL);
    return REDIS_OK;

werr:
    fclose(fp);
    unlink(tmpfile);
    redisLog(REDIS_WARNING,"Write error saving DB on disk: %s", strerror(errno));
    if (di) dictReleaseIterator(di);
    return REDIS_ERR;
}

int rdbSaveBackground(char *filename) {
    pid_t childpid;

495 496
    if (server.bgsavechildpid != -1 ||
        server.bgsavethread != (pthread_t) -1) return REDIS_ERR;
A
antirez 已提交
497

498
    server.dirty_before_bgsave = server.dirty;
A
antirez 已提交
499

500 501
    if (server.ds_enabled) {
        cacheForcePointInTime();
A
antirez 已提交
502
        return dsRdbSaveBackground(filename);
503 504
    }

505
    if ((childpid = fork()) == 0) {
A
antirez 已提交
506 507
        int retval;

508
        /* Child */
509 510
        if (server.ipfd > 0) close(server.ipfd);
        if (server.sofd > 0) close(server.sofd);
511
        retval = rdbSave(filename);
A
antirez 已提交
512
        _exit((retval == REDIS_OK) ? 0 : 1);
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686
    } else {
        /* Parent */
        if (childpid == -1) {
            redisLog(REDIS_WARNING,"Can't save in background: fork: %s",
                strerror(errno));
            return REDIS_ERR;
        }
        redisLog(REDIS_NOTICE,"Background saving started by pid %d",childpid);
        server.bgsavechildpid = childpid;
        updateDictResizePolicy();
        return REDIS_OK;
    }
    return REDIS_OK; /* unreached */
}

void rdbRemoveTempFile(pid_t childpid) {
    char tmpfile[256];

    snprintf(tmpfile,256,"temp-%d.rdb", (int) childpid);
    unlink(tmpfile);
}

int rdbLoadType(FILE *fp) {
    unsigned char type;
    if (fread(&type,1,1,fp) == 0) return -1;
    return type;
}

time_t rdbLoadTime(FILE *fp) {
    int32_t t32;
    if (fread(&t32,4,1,fp) == 0) return -1;
    return (time_t) t32;
}

/* Load an encoded length from the DB, see the REDIS_RDB_* defines on the top
 * of this file for a description of how this are stored on disk.
 *
 * isencoded is set to 1 if the readed length is not actually a length but
 * an "encoding type", check the above comments for more info */
uint32_t rdbLoadLen(FILE *fp, int *isencoded) {
    unsigned char buf[2];
    uint32_t len;
    int type;

    if (isencoded) *isencoded = 0;
    if (fread(buf,1,1,fp) == 0) return REDIS_RDB_LENERR;
    type = (buf[0]&0xC0)>>6;
    if (type == REDIS_RDB_6BITLEN) {
        /* Read a 6 bit len */
        return buf[0]&0x3F;
    } else if (type == REDIS_RDB_ENCVAL) {
        /* Read a 6 bit len encoding type */
        if (isencoded) *isencoded = 1;
        return buf[0]&0x3F;
    } else if (type == REDIS_RDB_14BITLEN) {
        /* Read a 14 bit len */
        if (fread(buf+1,1,1,fp) == 0) return REDIS_RDB_LENERR;
        return ((buf[0]&0x3F)<<8)|buf[1];
    } else {
        /* Read a 32 bit len */
        if (fread(&len,4,1,fp) == 0) return REDIS_RDB_LENERR;
        return ntohl(len);
    }
}

/* Load an integer-encoded object from file 'fp', with the specified
 * encoding type 'enctype'. If encode is true the function may return
 * an integer-encoded object as reply, otherwise the returned object
 * will always be encoded as a raw string. */
robj *rdbLoadIntegerObject(FILE *fp, int enctype, int encode) {
    unsigned char enc[4];
    long long val;

    if (enctype == REDIS_RDB_ENC_INT8) {
        if (fread(enc,1,1,fp) == 0) return NULL;
        val = (signed char)enc[0];
    } else if (enctype == REDIS_RDB_ENC_INT16) {
        uint16_t v;
        if (fread(enc,2,1,fp) == 0) return NULL;
        v = enc[0]|(enc[1]<<8);
        val = (int16_t)v;
    } else if (enctype == REDIS_RDB_ENC_INT32) {
        uint32_t v;
        if (fread(enc,4,1,fp) == 0) return NULL;
        v = enc[0]|(enc[1]<<8)|(enc[2]<<16)|(enc[3]<<24);
        val = (int32_t)v;
    } else {
        val = 0; /* anti-warning */
        redisPanic("Unknown RDB integer encoding type");
    }
    if (encode)
        return createStringObjectFromLongLong(val);
    else
        return createObject(REDIS_STRING,sdsfromlonglong(val));
}

robj *rdbLoadLzfStringObject(FILE*fp) {
    unsigned int len, clen;
    unsigned char *c = NULL;
    sds val = NULL;

    if ((clen = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;
    if ((len = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;
    if ((c = zmalloc(clen)) == NULL) goto err;
    if ((val = sdsnewlen(NULL,len)) == NULL) goto err;
    if (fread(c,clen,1,fp) == 0) goto err;
    if (lzf_decompress(c,clen,val,len) == 0) goto err;
    zfree(c);
    return createObject(REDIS_STRING,val);
err:
    zfree(c);
    sdsfree(val);
    return NULL;
}

robj *rdbGenericLoadStringObject(FILE*fp, int encode) {
    int isencoded;
    uint32_t len;
    sds val;

    len = rdbLoadLen(fp,&isencoded);
    if (isencoded) {
        switch(len) {
        case REDIS_RDB_ENC_INT8:
        case REDIS_RDB_ENC_INT16:
        case REDIS_RDB_ENC_INT32:
            return rdbLoadIntegerObject(fp,len,encode);
        case REDIS_RDB_ENC_LZF:
            return rdbLoadLzfStringObject(fp);
        default:
            redisPanic("Unknown RDB encoding type");
        }
    }

    if (len == REDIS_RDB_LENERR) return NULL;
    val = sdsnewlen(NULL,len);
    if (len && fread(val,len,1,fp) == 0) {
        sdsfree(val);
        return NULL;
    }
    return createObject(REDIS_STRING,val);
}

robj *rdbLoadStringObject(FILE *fp) {
    return rdbGenericLoadStringObject(fp,0);
}

robj *rdbLoadEncodedStringObject(FILE *fp) {
    return rdbGenericLoadStringObject(fp,1);
}

/* For information about double serialization check rdbSaveDoubleValue() */
int rdbLoadDoubleValue(FILE *fp, double *val) {
    char buf[128];
    unsigned char len;

    if (fread(&len,1,1,fp) == 0) return -1;
    switch(len) {
    case 255: *val = R_NegInf; return 0;
    case 254: *val = R_PosInf; return 0;
    case 253: *val = R_Nan; return 0;
    default:
        if (fread(buf,len,1,fp) == 0) return -1;
        buf[len] = '\0';
        sscanf(buf, "%lg", val);
        return 0;
    }
}

/* Load a Redis object of the specified type from the specified file.
 * On success a newly allocated object is returned, otherwise NULL. */
robj *rdbLoadObject(int type, FILE *fp) {
    robj *o, *ele, *dec;
    size_t len;
687
    unsigned int i;
688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728

    redisLog(REDIS_DEBUG,"LOADING OBJECT %d (at %d)\n",type,ftell(fp));
    if (type == REDIS_STRING) {
        /* Read string value */
        if ((o = rdbLoadEncodedStringObject(fp)) == NULL) return NULL;
        o = tryObjectEncoding(o);
    } else if (type == REDIS_LIST) {
        /* Read list value */
        if ((len = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;

        /* Use a real list when there are too many entries */
        if (len > server.list_max_ziplist_entries) {
            o = createListObject();
        } else {
            o = createZiplistObject();
        }

        /* Load every single element of the list */
        while(len--) {
            if ((ele = rdbLoadEncodedStringObject(fp)) == NULL) return NULL;

            /* If we are using a ziplist and the value is too big, convert
             * the object to a real list. */
            if (o->encoding == REDIS_ENCODING_ZIPLIST &&
                ele->encoding == REDIS_ENCODING_RAW &&
                sdslen(ele->ptr) > server.list_max_ziplist_value)
                    listTypeConvert(o,REDIS_ENCODING_LINKEDLIST);

            if (o->encoding == REDIS_ENCODING_ZIPLIST) {
                dec = getDecodedObject(ele);
                o->ptr = ziplistPush(o->ptr,dec->ptr,sdslen(dec->ptr),REDIS_TAIL);
                decrRefCount(dec);
                decrRefCount(ele);
            } else {
                ele = tryObjectEncoding(ele);
                listAddNodeTail(o->ptr,ele);
            }
        }
    } else if (type == REDIS_SET) {
        /* Read list/set value */
        if ((len = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;
729 730 731 732 733 734 735 736 737 738 739 740

        /* Use a regular set when there are too many entries. */
        if (len > server.set_max_intset_entries) {
            o = createSetObject();
            /* It's faster to expand the dict to the right size asap in order
             * to avoid rehashing */
            if (len > DICT_HT_INITIAL_SIZE)
                dictExpand(o->ptr,len);
        } else {
            o = createIntsetObject();
        }

741
        /* Load every single element of the list/set */
742 743
        for (i = 0; i < len; i++) {
            long long llval;
744 745
            if ((ele = rdbLoadEncodedStringObject(fp)) == NULL) return NULL;
            ele = tryObjectEncoding(ele);
746 747 748

            if (o->encoding == REDIS_ENCODING_INTSET) {
                /* Fetch integer value from element */
A
antirez 已提交
749
                if (isObjectRepresentableAsLongLong(ele,&llval) == REDIS_OK) {
750 751 752 753 754 755 756 757 758 759 760
                    o->ptr = intsetAdd(o->ptr,llval,NULL);
                } else {
                    setTypeConvert(o,REDIS_ENCODING_HT);
                    dictExpand(o->ptr,len);
                }
            }

            /* This will also be called when the set was just converted
             * to regular hashtable encoded set */
            if (o->encoding == REDIS_ENCODING_HT) {
                dictAdd((dict*)o->ptr,ele,NULL);
761 762
            } else {
                decrRefCount(ele);
763
            }
764 765 766 767 768 769 770 771 772 773 774 775
        }
    } else if (type == REDIS_ZSET) {
        /* Read list/set value */
        size_t zsetlen;
        zset *zs;

        if ((zsetlen = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;
        o = createZsetObject();
        zs = o->ptr;
        /* Load every single element of the list/set */
        while(zsetlen--) {
            robj *ele;
776 777
            double score;
            zskiplistNode *znode;
778 779 780

            if ((ele = rdbLoadEncodedStringObject(fp)) == NULL) return NULL;
            ele = tryObjectEncoding(ele);
781 782 783
            if (rdbLoadDoubleValue(fp,&score) == -1) return NULL;
            znode = zslInsert(zs->zsl,score,ele);
            dictAdd(zs->dict,ele,&znode->score);
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831
            incrRefCount(ele); /* added to skiplist */
        }
    } else if (type == REDIS_HASH) {
        size_t hashlen;

        if ((hashlen = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR) return NULL;
        o = createHashObject();
        /* Too many entries? Use an hash table. */
        if (hashlen > server.hash_max_zipmap_entries)
            convertToRealHash(o);
        /* Load every key/value, then set it into the zipmap or hash
         * table, as needed. */
        while(hashlen--) {
            robj *key, *val;

            if ((key = rdbLoadEncodedStringObject(fp)) == NULL) return NULL;
            if ((val = rdbLoadEncodedStringObject(fp)) == NULL) return NULL;
            /* If we are using a zipmap and there are too big values
             * the object is converted to real hash table encoding. */
            if (o->encoding != REDIS_ENCODING_HT &&
               ((key->encoding == REDIS_ENCODING_RAW &&
                sdslen(key->ptr) > server.hash_max_zipmap_value) ||
                (val->encoding == REDIS_ENCODING_RAW &&
                sdslen(val->ptr) > server.hash_max_zipmap_value)))
            {
                    convertToRealHash(o);
            }

            if (o->encoding == REDIS_ENCODING_ZIPMAP) {
                unsigned char *zm = o->ptr;
                robj *deckey, *decval;

                /* We need raw string objects to add them to the zipmap */
                deckey = getDecodedObject(key);
                decval = getDecodedObject(val);
                zm = zipmapSet(zm,deckey->ptr,sdslen(deckey->ptr),
                                  decval->ptr,sdslen(decval->ptr),NULL);
                o->ptr = zm;
                decrRefCount(deckey);
                decrRefCount(decval);
                decrRefCount(key);
                decrRefCount(val);
            } else {
                key = tryObjectEncoding(key);
                val = tryObjectEncoding(val);
                dictAdd((dict*)o->ptr,key,val);
            }
        }
832 833 834 835 836 837 838 839 840 841
    } else if (type == REDIS_HASH_ZIPMAP) {
        robj *aux = rdbLoadStringObject(fp);

        if (aux == NULL) return NULL;
        o = createHashObject();
        o->encoding = REDIS_ENCODING_ZIPMAP;
        o->ptr = zmalloc(sdslen(aux->ptr));
        memcpy(o->ptr,aux->ptr,sdslen(aux->ptr));
        decrRefCount(aux);
        /* FIXME: conver the object if needed */
842 843 844 845 846 847
    } else {
        redisPanic("Unknown object type");
    }
    return o;
}

848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872
/* Mark that we are loading in the global state and setup the fields
 * needed to provide loading stats. */
void startLoading(FILE *fp) {
    struct stat sb;

    /* Load the DB */
    server.loading = 1;
    server.loading_start_time = time(NULL);
    if (fstat(fileno(fp), &sb) == -1) {
        server.loading_total_bytes = 1; /* just to avoid division by zero */
    } else {
        server.loading_total_bytes = sb.st_size;
    }
}

/* Refresh the loading progress info */
void loadingProgress(off_t pos) {
    server.loading_loaded_bytes = pos;
}

/* Loading finished */
void stopLoading(void) {
    server.loading = 0;
}

873 874 875 876 877 878 879
int rdbLoad(char *filename) {
    FILE *fp;
    uint32_t dbid;
    int type, retval, rdbver;
    redisDb *db = server.db+0;
    char buf[1024];
    time_t expiretime, now = time(NULL);
880
    long loops = 0;
881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896

    fp = fopen(filename,"r");
    if (!fp) return REDIS_ERR;
    if (fread(buf,9,1,fp) == 0) goto eoferr;
    buf[9] = '\0';
    if (memcmp(buf,"REDIS",5) != 0) {
        fclose(fp);
        redisLog(REDIS_WARNING,"Wrong signature trying to load DB from file");
        return REDIS_ERR;
    }
    rdbver = atoi(buf+5);
    if (rdbver != 1) {
        fclose(fp);
        redisLog(REDIS_WARNING,"Can't handle RDB format version %d",rdbver);
        return REDIS_ERR;
    }
897 898

    startLoading(fp);
899 900 901
    while(1) {
        robj *key, *val;
        expiretime = -1;
902 903 904 905 906 907 908

        /* Serve the clients from time to time */
        if (!(loops++ % 1000)) {
            loadingProgress(ftello(fp));
            aeProcessEvents(server.el, AE_FILE_EVENTS|AE_DONT_WAIT);
        }

909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949
        /* Read type. */
        if ((type = rdbLoadType(fp)) == -1) goto eoferr;
        if (type == REDIS_EXPIRETIME) {
            if ((expiretime = rdbLoadTime(fp)) == -1) goto eoferr;
            /* We read the time so we need to read the object type again */
            if ((type = rdbLoadType(fp)) == -1) goto eoferr;
        }
        if (type == REDIS_EOF) break;
        /* Handle SELECT DB opcode as a special case */
        if (type == REDIS_SELECTDB) {
            if ((dbid = rdbLoadLen(fp,NULL)) == REDIS_RDB_LENERR)
                goto eoferr;
            if (dbid >= (unsigned)server.dbnum) {
                redisLog(REDIS_WARNING,"FATAL: Data file was created with a Redis server configured to handle more than %d databases. Exiting\n", server.dbnum);
                exit(1);
            }
            db = server.db+dbid;
            continue;
        }
        /* Read key */
        if ((key = rdbLoadStringObject(fp)) == NULL) goto eoferr;
        /* Read value */
        if ((val = rdbLoadObject(type,fp)) == NULL) goto eoferr;
        /* Check if the key already expired */
        if (expiretime != -1 && expiretime < now) {
            decrRefCount(key);
            decrRefCount(val);
            continue;
        }
        /* Add the new object in the hash table */
        retval = dbAdd(db,key,val);
        if (retval == REDIS_ERR) {
            redisLog(REDIS_WARNING,"Loading DB, duplicated key (%s) found! Unrecoverable error, exiting now.", key->ptr);
            exit(1);
        }
        /* Set the expire time if needed */
        if (expiretime != -1) setExpire(db,key,expiretime);

        decrRefCount(key);
    }
    fclose(fp);
950
    stopLoading();
951 952 953 954 955 956 957 958 959
    return REDIS_OK;

eoferr: /* unexpected end of file is handled here with a fatal exit */
    redisLog(REDIS_WARNING,"Short read or OOM loading DB. Unrecoverable error, aborting now.");
    exit(1);
    return REDIS_ERR; /* Just to avoid warning */
}

/* A background saving child (BGSAVE) terminated its work. Handle this. */
960
void backgroundSaveDoneHandler(int exitcode, int bysignal) {
961 962 963
    if (!bysignal && exitcode == 0) {
        redisLog(REDIS_NOTICE,
            "Background saving terminated with success");
964
        server.dirty = server.dirty - server.dirty_before_bgsave;
965 966 967 968 969
        server.lastsave = time(NULL);
    } else if (!bysignal && exitcode != 0) {
        redisLog(REDIS_WARNING, "Background saving error");
    } else {
        redisLog(REDIS_WARNING,
970
            "Background saving terminated by signal %d", bysignal);
971 972 973
        rdbRemoveTempFile(server.bgsavechildpid);
    }
    server.bgsavechildpid = -1;
974 975
    server.bgsavethread = (pthread_t) -1;
    server.bgsavethread_state = REDIS_BGSAVE_THREAD_UNACTIVE;
976 977 978 979
    /* Possibly there are slaves waiting for a BGSAVE in order to be served
     * (the first stage of SYNC is a bulk transfer of dump.rdb) */
    updateSlavesWaitingBgsave(exitcode == 0 ? REDIS_OK : REDIS_ERR);
}
980 981

void saveCommand(redisClient *c) {
A
antirez 已提交
982
    if (server.bgsavechildpid != -1 || server.bgsavethread != (pthread_t)-1) {
983 984 985 986 987 988 989 990 991 992 993
        addReplyError(c,"Background save already in progress");
        return;
    }
    if (rdbSave(server.dbfilename) == REDIS_OK) {
        addReply(c,shared.ok);
    } else {
        addReply(c,shared.err);
    }
}

void bgsaveCommand(redisClient *c) {
A
antirez 已提交
994
    if (server.bgsavechildpid != -1 || server.bgsavethread != (pthread_t)-1) {
995 996 997 998 999 1000 1001 1002 1003
        addReplyError(c,"Background save already in progress");
        return;
    }
    if (rdbSaveBackground(server.dbfilename) == REDIS_OK) {
        addReplyStatus(c,"Background saving started");
    } else {
        addReply(c,shared.err);
    }
}